The present invention relates generally to control systems, and more particularly to model predictive control employing novel techniques for optimizing throughput and yield.
Many processing applications include at least one batch process and at least one continuous process. In these types of systems, a major control issue is the optimization of throughput through the system. For instance, material through the continuous processes may continually flow through the process. However, material may only enter the batch process after a previous batch has been completely processed and may only exit once the batch process has been completed. As such, inventory levels at the continuous processes may remain generally constant whereas inventory levels at the batch processes may fluctuate over time, increasing at the beginning of a batch process and decreasing at the end of a batch process. Therefore, a main issue in the control of such systems may be to minimize the occurrence of supply constraints between the batch and continuous processes. In addition, a batch process of the system may affect a property of the product such that the property is enhanced the longer the batch process is allowed to continue. However, allowing the batch process to continue for longer periods of time may also decrease the overall system throughput. Therefore, a trade off may be made between overall system throughput and optimization of such properties, such as system yield.